论文部分内容阅读
随着微博等社交网络平台的兴起,民众参与线上生活、从中获取信息并相互交流的现象愈加普遍,分析社会网络中用户影响力等热点问题引起越来越广泛的关注.基于新浪微博提供的大规模数据集,利用用户所发微博产生的影响效果对用户影响力进行度量,将用户发微博时间、用户微博的转发、评论、点赞等交互行为与博文的内容相结合,分类提取特征,结合回归分析,预测用户影响力,并使用多视角堆模型融合方法改进预测精度.实验结果表明,提取用户行为与博文内容特征的方法可以有效预测用户影响力的变化,且随着特征数的增多、多视角堆模型融合方法的加入,预测准确度得到提升.
With the rise of social networking platforms such as Weibo, the populace’s participation in online life is getting more and more widespread, and the analysis of hot issues such as user influence in social networks has drawn more and more attention. Based on Sina Weibo Provides a large-scale data set that uses the effect of microblogging made by the user to measure the influence of the user, combines the user’s microblogging time, the user’s microblogging forwarding, comments, likes and other interactive activities with the content of the blog post , And the feature of classification was extracted.The regression analysis was used to predict the influence of users and the prediction accuracy was improved by using multi-view heap fusion method.The experimental results show that the method of extracting user behavior and content features can effectively predict the change of user’s influence, With the increase of the number of features and the integration of multi-view heap model fusion method, the prediction accuracy is improved.